I think the blackout is counter productive to continue. The idea was to try and convince reddit to change their minds, but as we can see now that is simply not going to happen. you can't convince someone who is unreasonable with logic. The way it has gone is that there has been a large influx of people trying things like kbin and lemmy, and the experience has not been smooth. lemmy.ml being immediately overloaded, and continuing to be difficult to federate with reliably is a problem, and not one just for that instance. It creates a sour view of the network (especially with other instances running into issues that seem to be scaling issues with the lemmy software). I think kbin has had some scaling issues that required using a large cdn to even try to cope with. The problem there is that if you look at the traffic of some of the largest federated reddit replacements, it is almost nothing in the grand scheme of things. The number of communities is very small, and those that exist have thread counts in the double digits at best. Most of that is due to the hyper isolation and fragmentation of instances due to federation issues at scale.
In the end I think every sub should call it a day on blackout. Go back to normal and make it clear that reddit hasn't won, you are just regrouping. Then give the software devs a bit more time to work before the 3rd party apps go offline for the final time. use the initial blackout as a scream test, and wait for reddit to disable themselves by revoking api access. Hopefully by then many of the larger issues with reddit like federated alternatives can be resolved, or at the very least minimized, so the transition away from reddit can take place quickly and easily.
it is easy enough. Simply run the playbook again. well, git pull the ansible playbook again and then run it. alternatively you can just use docker compose now on your lemmy server. I made some aliases on my lemmy instance based on what i use elsewhere. I think I got them from a linuxserver.io tutorial ages ago. you will need to adjust the container versions for this to be viable as the version is hardcoded and they only have a "latest" tag for arm.